The purpose of this application is to develop structure-activity models of the NCI/NTP and other carcinogenesis bioassays. Through these models it would become possible to predict the probability of carcinogenicity of untested compounds. The application would be to those chemicals that have not as yet and probably never will be tested already in current commerce, as well as those being introduced at the rate of approximately 1,000 per year into commerce. Other applications exist at the developmental stage to enhance the safety of laboratory and production personnel, and to minimize the use of animals in research and testing. During Phase I it is expected that the feasibility of this approach can be demonstrated. During Phase II a refinement of the approach and commercialization would be expected. The methodology proposed in this application consists mostly of collection of results of carcinogenesis bioassays, their critical evaluation, and modeling of the assembled data bases by means of conventional multivariate statistical methodology. The technological innovations expected from this project would be the ability to predict which chemicals are likely to be carcinogens on the basis of structure-activity equations to a degree not heretofore possible. The commercial applications would be to those compounds in the Toxic Substances Control Act inventory which have not been tested, to chemicals currently being developed, and as a tool in its software implementation.